This document is a 5 page exam for a low power IC design course consisting of 9 questions. Question 1 asks to calculate switching activity and transition density for an AND gate under different input conditions. Question 2 asks about optimizing supply voltages for a system to minimize power without increasing latency or developing a time multiplexed bus. Question 3 is about short circuit current variations in a CMOS inverter. Question 4 is about software level power analysis of a sample program. Question 5 asks to design full adder circuits using different methods and compare their efficiency. Question 6 involves a BDD diagram and power saver modes. Question 7 involves FIR filter structures and estimating power savings. Question 8 is about designing a 6T SRAM cell with reduced bit
This document is a 5 page exam for a low power IC design course consisting of 9 questions. Question 1 asks to calculate switching activity and transition density for an AND gate under different input conditions. Question 2 asks about optimizing supply voltages for a system to minimize power without increasing latency or developing a time multiplexed bus. Question 3 is about short circuit current variations in a CMOS inverter. Question 4 is about software level power analysis of a sample program. Question 5 asks to design full adder circuits using different methods and compare their efficiency. Question 6 involves a BDD diagram and power saver modes. Question 7 involves FIR filter structures and estimating power savings. Question 8 is about designing a 6T SRAM cell with reduced bit
This document is a 5 page exam for a low power IC design course consisting of 9 questions. Question 1 asks to calculate switching activity and transition density for an AND gate under different input conditions. Question 2 asks about optimizing supply voltages for a system to minimize power without increasing latency or developing a time multiplexed bus. Question 3 is about short circuit current variations in a CMOS inverter. Question 4 is about software level power analysis of a sample program. Question 5 asks to design full adder circuits using different methods and compare their efficiency. Question 6 involves a BDD diagram and power saver modes. Question 7 involves FIR filter structures and estimating power savings. Question 8 is about designing a 6T SRAM cell with reduced bit
This document is a 5 page exam for a low power IC design course consisting of 9 questions. Question 1 asks to calculate switching activity and transition density for an AND gate under different input conditions. Question 2 asks about optimizing supply voltages for a system to minimize power without increasing latency or developing a time multiplexed bus. Question 3 is about short circuit current variations in a CMOS inverter. Question 4 is about software level power analysis of a sample program. Question 5 asks to design full adder circuits using different methods and compare their efficiency. Question 6 involves a BDD diagram and power saver modes. Question 7 involves FIR filter structures and estimating power savings. Question 8 is about designing a 6T SRAM cell with reduced bit
Max.Marks:100 Answer ALL Questions 1. Compute the switching activity () and the transition density (D) at the output of a AND gate with 2 inputs a and b for the following cases, given that they are implemented using static CMOS style. Also assume that the total possible events for each of the following cases occur over a reference time frame of 1 sec. a) Uncorrelated inputs with uniform probability distribution. b) Temporal correlation in the inputs: Every 0 applied to input a is immediately followed by a 1, while every 1 applied to input b is immediately followed by a 0. c) Spatial correlation among inputs: only patterns 00 and 11 can be applied to the inputs and both are equally likely. d) Spatio-temporal correlation: changes exactly when b changes value. [15] 2.(a) i. Consider the system shown in the fig below, where all blocks are assumed to be identical. Let us assume that each block in the system has a critical path of 1 time unit when the system is operated with a supply voltage of 3V. The threshold voltage of the technology, V th , is known to be 0.5V. Assume a maximum of 3 supply voltage to be available. Also assume that the minimum supply voltage to be 1.33V. Use multiple supply voltage to obtain the least power consumption without altering the system latency of 6 u.t. Neglect loading effects.
[8]
Page 2 of 5
ii) Develop hardware for time multiplexed bus sharing for the fig shown below.
[2]
[OR] [10] 2.(b) Consider the following different implementations of the logic functions=(abc+def) using the characteristics cells of a low area 0.8m library, listed in the table-1 below. i. Implementation 1: Using NOR and inverter cells. ii. Implementation 2: Using NOR, NAND and inverter cells. Assume a load capacitance of 0.2 pf. Determine which of the above implementations is power efficient, by computing the power cost, if the inputs are uniformly distributed and uncorrelated. Table-1: Gate type Output capacitance (fF) Input capacitance (fF) Inverter 35 13 NAND2 60 13 NAND3 65 13 NOR2 62 13 NOR3 69 13
3. Get the transfer characteristics of a CMOS inverter and find the short circuit current and sketch the dependencies of short-circuit current on the following parameters i) SSC variations with output load capacitance and how to get the optimized design. ii) SSC variations with different input slopes and how to get the optimized design. [10] Page 3 of 5
4. Mention the significance of software level power analysis. Consider a sample program execution profile described in the table below. The base current cost and the number of cycles for each instruction of an INTEL 486DX2 processor is listed alongside. The measured current cost for this sequence is 332.8 mA. Hence determine the following: a) Base cost energy and power cost for the program. b) Energy cost and power cost due to the circuit state overhead. Assume that the processor operates with a supply voltage of 3.3V with a clock of 80Mhz. Further assume that there are no instruction stalls and cache misses. The sample program goes as follows: ADD DLOAD LOAD; MULT ADD [10]
Instruction Base Cost(PJ) Circuit State effect (PJ) Cycles Load Dload Add Mult Load; Add Load ;Mult LOAD DLAOD ADD MULT LOAD;ADD LOAD;MULT 1.98 2.37 0.99 1.19 2.10 2.25 0.13
5. Design a Low Power Full adder circuit in the transistor level using the following methods: a) XOR-XOR Based Full adder b) XNOR-XNOR based Full adder. Compare their efficiency with conventional full adders in the aspect of number of transistor used, delay and power. [10] 6. a) Develop a BDD diagram for the function was expressed as y = x3 + x1 (x2 + x3). b) Describe the three power saver modes in the system level. [10] 7.(a) Draw the direct FORM-I structure of a 8-tap FIR filter. Assume the propagation delay of a multiplier to be thrice that of an adder. For power calculation purposes, further assume that the total capacitance of a multiplier is 10 times that of an adder. Draw an alternate FIR filter structure whose critical path delay is exactly half of the previous version. If the latter filter structure is to be operated at the same sampling rate as that of the former, then estimate the following. i. the supply voltage for the alternate filter structure and ii. the savings in power consumptions expressed as a percentage of the original filter. Assume Vt=0.5V and V dd for the original filter is 4V. [15] [OR]
7.(b) i. Consider a static 2-input NOR gate and assume that only one input transition is possible during a clock cycle and also assume that the inputs to the NOR gate have a uniform input distribution of high and low levels. This means that the four possible states for inputs A and B (00, 01, 10, 11) are equally likely. I. Find the probability that the output is 0. II. Find the probability that the output is 1. III. Find the 0 to 1transition probability. Draw the state transition diagram and annotate with transition probability. ii. Consider the example in which the multiplication by a constant is optimized by decomposing into a shift and add operation using a CSD representation. Consider the two topology shown below for the function using the Associativity and commutativity law Transformation, comment which topology give me the less number of transition probability. [7]
[8] Page 5 of 5
Fig (a) Fig (b) 8. Design a 6-T SRAM cell with reduced bit line swing and pulsed word line and discuss various way of reducing power dissipation in the cell. [10] 9. Derive the expression for energy dissipation in a transistor channel using an RC model. Also highlight the framework for an energy recovery circuit design. [10]